|8:00am||Continental Breakfast - Thank you IEEE 1149.4|
|8:30am||Arrival and Introductions||Adam Cron|
|9:00am||Approval of October, 1998 Minutes||Adam Cron|
|9:30am||Standard Status||Adam Cron|
|10:00am||New Officers||Adam Cron|
|10:45am||Panasonic Application||Katsuhiro Hirayama|
|11:15am||Marketing New Standard||Steve Sunter|
|12:00pm||Adjourn for Lunch|
|2:00pm||Vth Topic||Steve Sunter|
|2:45pm||New Technical Papers||Adam Cron|
|3:00pm||Break - Thank you IEEE 1149.4|
- What Is Known Today (Topological)
- What Is To Be Learned (Parametric/Error)
| Ken Parker
|5:30pm||1149.1 Status||Adam Ley|
|5:45pm||Next Meeting||Adam Osseiran|
|Working Group Members||31|
|Total Subscribers on "esd" reflector||376|
|Companies and Other Institutions Participating||~300|
|John Andrews||National Semiconductor|
|William Aronson||National Semiconductor|
|Jeff Butler||National Semiconductor|
|Tapan J. Chakraberty||Lucent Technology|
|Chen Huan Chiang||Lucent Technology|
|Frans de Jong||Philips Research|
|Duy Le||Lucent Technology|
|Keith (Katsuhiro) Hirayama||Panasonic|
|Keith Lofstrom||KLIC, SiidTech|
|John McDermid||Agilent Technologies|
|Elbert Nhan||Johns Hopkins University|
|Hideo Okawara||Agilent Technologies|
|Adam Osseiran||Fluence Technology|
|Ken Parker||Agilent Technologies|
|Mani Soma||University of Washington|
|Martin Viktil||SINTEF Electroncs and Cybernetics|
|Brian Wilkins||Southampton University|
Adam C. reminded everyone of the rules for becoming a Dot-4 member. He said the requirement is that a prospective Dot-4 member must attend at least two of three consecutive meetings and over a two-year span. Steve and Mani said only two out of three consecutive meetings are required.
Adam C. said we might be able to finish everything on the agenda by noon so some people could catch their flights home.
Adam C. solicited sponsors for future meetings. He reiterated that sponsors get their names on the meeting agenda and in the Working Group (WG) minutes that are posted on the 1149.4 web site. Sponsors also receive thanks from the working group.
There are a total of 564 interested individuals in the Dot-4 database. Currently, there are two email reflectors: One for the WG and the other for non-WG members. Officially, WG members number 31 as of October 1999. Adam C. will include in the Dot-4 WG reflector the 31 WG members, chairs of other working groups and selected international people. It is supposed to be private area.
Adam C. showed the attendee list. He announced that Brian Wilkins did not come to this meeting and has essentially retired from the WG.
Mani Soma motioned to approve the October 1998 WG meeting minutes. John Andrews seconded. Unanimous approval. Motion carried.
Steve Sunter asked if anybody read the minutes. They are posted on the web.
Adam C. thanked the group for making approval possible. He especially thanked Brian for his tenaciousness and asking lots of questions and eventually getting the answers to those questions.
John Andrews suggested sending an email to Brian to thank him. Adam C. said the IEEE would present each of the four officers with a plaque. Steve said we should send Brian a note anyway.
John Andrews motioned to send an email to Brian Wilkins. Frans de Jong seconded. Unanimous approval. Motion carried.
On p. 16, Definition 3.1.7 is one place where it fits the bill. Ken pointed to this saying it is permission language. Steve said it is a functional description. Ken said we need to take it out too.
Steve Sunter and Ken Parker will identify in the minutes and the document where neither "input" nor "output" word was used.
Ken Parker motioned to remove the word "output" from Rule 7.3.1 (b). Mani seconded. Unanimous approval. Motion carried.
Steve commented that in the past it was usually 50-50 voting on anything but now it is often unanimous. Adam C. said there have been about 17-20 members that were regular at every meeting. Steve said we rarely accepted anything that was only halfway agreed on.
Adam C. said a few words about Standard status. In 2 weeks, the IEEE will be done with the editing and we will have the 107-page document on hand. Steve wanted to review the copy (proofread) after being edited prior to dissemination. Steve said there have been a lot of changes since we submitted the Draft for approval.
Ken asked if the IEEE would put the final copy in some weird format that nobody can edit it? Adam C. said there would be a new editor who may use a different word processor.
Adam O. said he himself is not very active. He was impressed with the Tiger Team’s work. He would like to keep doing the same things Adam C. has done with the interest of keeping it going. He commented that he did not know what lies ahead but would accept any challenges. Mani added that Adam O. pushed for Dot-4 in Europe and held various meetings on the continent. In short, Adam O. is pretty active in Europe. Ken said that in the offseason it would probably be cheaper to go to Switzerland anyway. We can hold meetings either in Europe or the U.S. and will keep the annual ITC meeting. Tony announced he had dropped out of consideration for chair. Adam C. said he did not sense any objection among the meeting attendees to Adam O.
Steve motioned that Adam Osseiran be the new 1149.4 Chair. Mani seconded. Unanimous approval. Motion carried.
Vice Chair: Steve indicated that he would like to stay on. Adam C. asked if there were any other candidates. There were no objections to Steve staying on as Vice Chair. Adam Sheppard was recruited to be chair but has since moved on to another company.
Mani motioned that Steve Sunter remains as Vice Chair. John McDermott seconded. Unanimous approval. Motion carried.
Editor: The main responsibility of the editor is to just "tweak" the document. But with BSDL, it may involve a lot more BSDL work and not just fine-tuning the document. Initially, BSDL will be the only significant work. Adam O. noted that Dot-6 might be started soon. So BSDL may not even get included in Dot-4. Steve concurred that BSDL could very well be a separate item and that we could defer the editorship for at least a year. Adam C. said he wanted to be the Editor since there is not a whole lot of BSDL language involved. He will do it as long as there are not a lot of work or a lot of changes. Keith said the Editor’s tasks could be onerous like Adam Ley’s of the Dot-1 working group. Ken pointed out that Adam L. essentially turned an unusable document into a usable one.
Steve Sunter motioned that Adam Cron be the new Editor. Keith Lofstrom seconded. Unanimous approval. Motion carried.
Secretary: Elbert said anyone in the WG is welcome to take his job, which consists of taking minutes at WG meetings. Other tasks include casting votes on motions and issues and participating in technical discussion if necessary. Recording what was said in meetings requires a certain level of concentration and understanding of the issues discussed. The Secretary would edit the meeting notes and minutes and send them along with the viewgraphs and relevant papers used in the WG meetings to the Chair for review and posting on the Dot-4 website. Keith Lofstrom commented that with the responsibilities of the Secretary just described, it probably discouraged most of the people from considering to take for the position.
Ken Parker motioned that Elbert Nhan remain as Secretary. Adam Osseiran seconded. Unanimous approval. Motion carried.
INTEST discussion was officially tabled and the WG moved on to another topic under BSDL-A.
For the following discussions, refer to Steve’s Viewgraphs VG1, VG2, VG3, and VG4. We needed and now have an official IEEE 1149.4 web site. We can increase the amount of information that gets posted there. Right now, we only have the minutes and reports of the test chips posted. Ken asked how we could determine if a chip is Dot-4 compliant. We are getting into tricky grounds here. We can say the chips are "claimed" to be Dot-4 compliant. We do not limit the label to just off-the-shelf components but also to custom ICs. Frans said there should be a bullet on Viewgraph VG1 for initiating the new people to the Standard and maybe some tutorial materials. Ken said along with the standard IC, there are people out there providing testing services and developing software, etc. Adam C. said we have to be careful when approving products. One thing we could do is put a disclaimer on the web site saying we are not necessarily sanctioning any of the products or promoting one over the others. Steve said we should put a link on the web site that where a list of companies with products and services can be found. The key point is to convey to the public on our web site that the industry is embracing and receptive to the Standard, and that more and more people are adopting the Dot-4. Carl Bernhart suggested that we say everything is "claimed compliant" so we would not have to pass any judgment. Steve said we should collect all the example products we can find that make use of Dot-4. We can also have a FAQs section. Adam C. and Carl both raised the possibility of putting a form on the web site with which people can request info or have their ICs or test services listed. Adam C. said we are an IEEE web site and should be careful not to let it be commercialized. The homepage is strictly a service to IEEE members. Adam C. said that in the past companies had requested to have their wares shown at IEEE meetings, and we should avoid that. We can keep adding reports and results to the web site as they become available. We should compile and post a list of published papers that have anything at all to do with Dot-4. But we cannot post IEEE papers without permission of their authors. The ITC explicitly makes you sign to relinquish ownership of your own papers. The IEEE owns those papers that get published in the proceedings of conferences and symposia it sponsors. Only with IEEE approval can papers be posted on our web site. Keith Lofstrom said he put papers and links on his own web site. Steve said we should post a few published books on Dot-4 (one by Ken and two others). Steve suggested a section on tutorial. Ken asked if we could get statistics on the number of hits generated. Adam C. said he tried but it did not quite work. Tony Suto said that for the purpose of marketing Dot-4, we could provide the reasons in a concise form why people should use Dot-4 and how they could benefit from it. Steve agreed. Adam C. said right now there is a charter statement on the web site. Steve said we could replace the statement with the benefits of Dot-4 as suggested by Tony.
Carl asked if there should be an email address for people to send comments to. Adam C. said they could send them to the private reflector. Carl said he was considering putting this web site on his unofficial web site at work. We need to filter comments and worthwhile comments from the rest and publish them along with the answers to them. Carl said that we need to explicitly tell people to direct questions of interpretation of the Standard to the current Chair. Keith said we could very well get bizarre questions and comments on the web site. If there are questions about a specific company’s products, we should forward them to that company’s site.
Steve said we should reference Dot-4 in all publications (Viewgraph VG3). Keith said we might want to convert to .com status. $70 for 2 years is the cost. But Carl said we would still have to pay to maintain the site. Keith said we could get companies like Panasonic or Phillips to sponsor it. Steve said we do not want to put this web site on some company’s web site. We could indicate explicitly that even though this is a .com site, it really belongs to the IEEE.
Finally, Steve said the cost of Dot-4 is high compared to its value (See Viewgraph VG4). You are asking analog guys to put 6 extra pins on an IC, and that is asking a lot. It is difficult to decrease the cost of Dot-4; so we need to increase the value of Dot-4. We, as a group, can work on maximizing the value but we cannot do a whole lot as far as reducing the cost of implementing the Standard is concerned. Adam O. broached the system-on-chip (SOC) working group. Steve went to their meetings in the past and made sure they did not come up with anything new to Dot-4. All the SOC working group wanted is for Dot-4 to be hierarchical, and that is why we have worked it into the Standard. Tapan Chakraboniy mentioned the P1500 Working Group. Mani asked who in Dot-4 goes to their meetings. Only Tapan and Adam O. do. Mani said we should have somebody from Dot-4 go to P1500 meetings and pitch for Dot-4. Their meeting is sometime in November.
Steve said wider usage of Dot-4, such as in telephone and Internet, will result in increased value of the Standard. Credibility and value also increase with each new user, product, capability, and IC. If many people contribute, growth can be exponential. Ken said people tend to remain quiet until they are comfortable with it and then say, "Oh yeah, we have done that 2 years ago."
Keith suggested internationally, marketing should be promoted. Steve agreed, saying that maybe we could put in various languages the benefits section on the web site so people from foreign countries could quickly grasp the benefits of Dot-4 when visiting the site. Adam O. said introductory materials could also be translated into various languages on the web site. Adam C. said he would try to have that done on the web site by the end of year. Anybody that prepares introductory materials should refer to the web site. Carl said we need to put together a quality presentation for marketing purpose. Ken suggested the local IEEE chapters could have a little discussion of Dot-4. Mani has Dot-4 slides that he will forward to Adam C. They are on his web site. Keith said we should use PowerPoint whenever possible to make slides.
The WG discussed the dissemination and marketing of Dot-4. The Dot-4 web site serves as a starting point. At present, the site is relatively "bland," but appropriate educational and relevant materials, such as a list of published papers, FAQs and Dot-4 benefits sections, will be posted on the site. There was talk of turning the site into a .com one with sponsorship from member companies. The aim should be to bring as much exposure to Dot-4 as possible, both in the U.S. and internationally. This can be accomplished by having Dot-4 posted in various languages and participating in other working groups and meetings.
In practice, one should select a reasonable value for Vth, but not exactly halfway between Vmax and Vmin because of the possibility of the pin in a metastable state. After some debating on whether a variable Vth should be mandated in Dot-4, the WG decided that it should not be at the risk of Dot-4 being too restrictive to designers. Therefore, no changes were proposed to the Vth rules.
Permission (q) says, "additional switches may be included to support additional instructions and modes of operation. Any such additional switches shall not interfere with the operation of any of the provisions of this standard or of IEEE Std 1149.1." These switches do not need to be documented.
In 6.5 (Partitioned internal test bus structure) with regard to ABUS extension, we have to know which function pin is associated with each control cell.
In Chapter 7 (Boundary Scan Register), it does not make a difference to BSDL whether or not a pin is digital. Rather, the important thing is the available resources. Steve asked if someone had a digital pin but wanted analog parametric capabilities with it, how do would he/she do it and document it? Ken said that from the software standpoint, for INTEST of Dot-4 for ABM, the pin is continuously connected to core (forced to drive in both directions) but for INTEST of Dot-1, the core is disconnected. If an ABM is used for INTEST, then we cannot just drive the core but also in both directions. Ken said in Fig. 10 of p. 35, we are in a murky area. Will INTEST have any meaning in that situation? Again, in INTEST, for Dot-4, the pin is continuously connected to core. How do we document it, Steve asked, if we allowed both the ABM and the DBM to connect to it? A quick solution is to put the digital part on the core side of the CD switch. A DBM is between the analog and the digital and is documented.
INTEST of Dot-4: we cannot drive to logic 1 or 0. Refer Viewgraph VG7. Keith said that if a behavior is observed all the time, then we do not document it. However, if it only happens sometimes, then we should document it.
P. 33, 22.214.171.124, Rule (b), there is not a way to distinguish between analog and digital, Ken said. We will have to identify analog and digital pins in BSDL.
Adam C. said if a pin has an ABM on it, it must be an analog pin. But Steve pointed out that we can have a digital pin that is ECL and we would like to measure the pull-up resistor value with an ABM. Keith said that we would need a third cell that uses AB1 and AB2 for other purposes.
Carl commented that we could have an analog ABM on the pin side and a DBM on the other side of the CD switch. Tony said in 126.96.36.199 of Version D26, there is a requirement for each analog-digital interface.
The fundamental question is whether we can put DBM and ABM in series and still have to document it? Carl said we perhaps should consider putting this in the Standard in the future. A semantic distinction between analog and digital INTEST would be a topic of interest.
Ken summarized: if you put an ABM on a digital pin and if you want to support INTEST, then you would need a DBM. Essentially, the only situation in which an ABM is found in series to DBM is when INTEST is involved. Steve thought the circuit on Viewgraph VG7 is a nice circuit. John McDermid said that with this circuit we couldn’t say we are running a digital or an analog INTEST. Why would it matter? We do not have enough information to say either way. Ken added that for input pins where INTEST is allowed, we have the same problem.
188.8.131.52 Rule (b) says that when INTEST instruction is selected, all analog pins shall be connected to the analog core. So, no rules are violated since the analog core is a dead-end piece of wire in the case depicted in Viewgraph VG7. Keith said he rather add a bit than chop the ABM in halves.
At this point, with Adam O. and Frans leaving to catch their flights, Keith proposed to debate this topic over email.
Ken continued the BSDL-A discussion by going through the Draft chapter by chapter to see if anything needs to be reviewed and discussed. Wherever possible, 2 differential signals should be connected to 2 ABMs. There will be impacts to BSDL because modifications are required. Right now in Dot-1, rule (b) is implicit on p. 59. As things now stand, 3 ABMs are required (2 observe only; 1 observe and control). Rules (c) and (d) require some support on BSDL that will be different from what we have today. Adam C. said some of the changes proposed to Dot-1 got dropped in the current version of Dot-1, and the entire page could be very messy. Ken said we would have to work on that page heavily.
On p. 62, in 7.3.1, the discussion here is almost identical to that for TBIC. We have to know the thresholds, Vhs and Vls. But we do not have the extension stuff and therefore it is a little easier. Another question is do we need to know what Vg is. John M. said that we need to know what pin Vg is referred to. It could be associated with the "source" pin where the signal originated from for each function pin. It is really a per-pin specification. Hence, we could just delete the word "output" from Rule (j). Rule (j) is redundant.
Residual elements: Just mark it as an agenda item? Steve asked if we have a Vh that is approximately close to Vdd. The impedance of those Vdd, Vss switches are covered in the Parameter section.
Steve said 7.3.1 is just like the TBIC case. We need to know if a buffer is bidirectional. If it is a buffer we are talking about here, we need to know its gain. Ken does not think w have to know that since the voltage/current capability is optional.
Other topics of discussion included the following: BSDL has provision for denoting the version of test software and Permission (g) is a "don’t care".
In 7.4 (Differential Analog Boundary Modules): Ken said to just mark it down as a major topic as a continuation of the other major projects.
Chapter 8: No changes are needed.
In 9.3.1 (Rcom), residuals always have to be documented, Steve said. Ken said we are getting into an area where things have to be documented. Right now there are no provisions in BSDL to describe the parameters of residual elements. Ken predicted that some of the new items would have to go in there. There are no clear answers. Steve said the minimum documentation is whether we can do an interconnect test on a pin. On p. 115, the text between Table 11 and 10.1 (Residual Elements) will have to be addressed. The TBIC is another topic that requires review and discussion. 10.2.2 indicates what we are supposed to document but it does not tell what has to be in BSDL (refer to Viewgraph VG8).
Rule (b) is a subset of Rule (a). Rules (a) through (d) are already included in the current version of BSDL.
Should Rule (e) be part of BSDL or can just be included in a databook somewhere? In other words, do we need a BSDL description for it? Perhaps it does not need to be documented in BSDL. Unless it is useful at the board level or above, then we would do something with it. Steve said maybe we should leave it alone for now, and if the market demanded it, we will put it in. Items (ii) to (iv) are covered. Steve said Item (iii) may not be documented anywhere in the Dot-1 document. It is important for simple interconnect test.
Rule (f): We have already covered the TBIC. For each pin of the ATAP, there should be a model table for it. Residuals only apply to function pins. The purpose of documenting these residuals is to aid testing speed and effectiveness but their values do not have to be known. These residuals will help in measurements but the question is whether we need to document impedance, capacitances of AB1, AB2, etc. John M. suggested that only the exceptions be documented. Before the WG digressed from this topic, it was tabled for further discussion via email.
Tony said we probably should not use hard limits but rather approximate values. Steve said all analog values have tolerance to them. Tony commented that leakages are usually a function of temperature and other environmental parameters. On p. 114, Items (2) through (5) give the minimums and maximums for various residual parameters of interest.
Ken asked if signs are taken into account in the minimum and maximum values. That is another thing to fix, Keith said.
Ken still was not quite sure if everything we talked about has to be in there for the software to work. John M. said we need to make the distinction between things needed for test generation and things that are for documentation but not necessary for test generation. Carl said we might need a multi-level type of cell descriptions, perhaps something like a table of technology.
Carl raised the question of whether a digital chip could be claimed Dot-4 compliant if it had differential cells in it. Steve pointed out that we could not claim Dot-4 compliance without an AT1 and an AT2. In that situation, one cannot really claim compliance to either Dot-1 or Dot-4. A finger should be pointed to Dot-1 that does not cover every imaginable case out there.
SUMMARY: The BSDL software implications were discussed. The entire Dot-4 draft was reviewed to make sure Dot-4 will work with BSDL. Issues like the need to document switches in the software, TBIC, and residual elements were examined. The topics of INTEST and residuals were delegated to email discussion.
Basically, Dot-1 came out with a draft and went to ballot. The one that Adam C. has is a substantially complete copy. The Draft is complete on a technical basis. Balloting is currently out. A balloting group has to be formed. Adam L. expected to receive comments from people in another 6-9 months. For anyone interested in hearing the technical details, Adam L. summarized:
With regard to the Whet-cells, provisions of digital facility to the analog pins were left out. There is a clear split. Either something is a Dot-1or Dot-4 chip. Therefore, no digital test facility at the analog pins was recommended. Brian Wilkins did not request any changes to Dot-1. Ken said that Dot-1 says "thou shall not touch analog pins." Steve asked whether you could say if a chip is Dot-4 compliant that it cannot be Dot-1 compliant. Mani said the major selling point of Dot-4 is that something is automatically Dot-1 compliant when it is Dot-4 compliant. We would contradict ourselves if we claimed it was not true. Ken said Rules (a) and (b) on p. 22 of the Dot-4 Draft say a chip is Dot-1 compliant if it is Dot-4 compliant. Adam O. asked that for systems-on-chip (SOCs), what if we had mixed-compliances of chips on a superchip, to what would the SOC be compliant then? Basically, not all Dot-1 chips are necessarily Dot-4 compliant, although they could be. Ken said we should leave it to test engineer to resolve that issue. Anyhow, the Whet-cells are now not mandated in Dot-1.
From a practical standpoint, we were on a long journey and went in full circles and now decided to go with the status quo. Ken said we went in full circles a lot and Brian used to agonize over this for a long time. Adam L. said if someone had a 24-bit digital-to-analog converter, Dot-1 says you should have 24 boundary cells, but not Dot-4. That should be one of those FAQ questions posted on the web.
A few things have been added to the new Dot-1draft. Among those are as follows: 1) SAMPLE is now "divorced" from the PRELOAD but both are still mandatory. 2) It is now mandatory that the "All 0s" op code be discarded. The recommendation is not to use that op code during the test mode to prevent accidental loading. There were minor changes to the language in the Draft. Keepers were added to BSDL. Also, Bill Bruce pointed out that it is not required to turn off the keeper in EXTEST or High-Z. Ken, Adam, Bill, Colin Maunders were involved in this issue. The key point is that Dot-4 is moving forward instead of staying put. Adam L. had a 6-page summary of the key changes he has just discussed that he could send out to interested people. The changes are also mentioned in the introductory paragraph of the Dot-1 draft. Adam C. added that one thing out of the Dot-1 meeting is that the balloting group is being reformed. If you want to vote, you would have to pay $10 to get back in. Carl said you would have to be a member of the IEEE Standards Association (SA) if you want to vote in the next Dot-4 ballot.
Adam Ley briefed the WG on the latest developments in Dot-1. Basically, the most significant development is that the Whet-cell is not mandated in Dot-1, and Dot-1 treats analog pins as before – one shall not concern with them. The implication is that analog pins are in Dot-4’s domain. Other changes included the requirement that SAMPLE be separated from PRELOAD but both are still mandatory, and that the "All 0s" op code be eliminated. For a complete detailed list of the latest changes, one can visit the Dot-1 homepage. The latest Dot-1 draft is in balloting.
1. VLSI Test Symposium (VTS) in Montreal, Canada in early May or late April (14).
2. European Test Workshop (ETW) in Lisbon, Portugal in June (7).
Some in the WG felt June is right time frame. We could even submit papers to the ETW and have a session on Dot-4 in Europe. What about sponsors? Agilent? LogicVision? It is still six months away; we can ask for sponsorship of the WG meeting later.